what changed after the inference shift for Qwen

[macropilot]

Posting this in case it saves someone else a little time.

If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Would rather learn from your scars than collect my own all at once.

[sparrowshed]

Posting this in case it saves someone else a little time.

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with FLUX and compare from there.

[macropilot]

What exact setup are you running? Do you see the same thing with a smaller batch or shorter prompt?

[localpatch78]

worth checking the driver first.

[tensorfield88]

I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at quant drift.

[signalcraft14]

that fixed it for me too.

[prompttinker44]

I'd approach it in a boring order. If you're on llama.cpp, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[routearch]

A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at speculative decoding. Not glamorous advice, but repeatable baselines save a lot of time.

[kernelmike]

A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at speculative decoding. Not

I'd approach it in a boring order. On A6000, the easiest win is often to back off the aggressive settings and see whether the problem survives. Not glamorous advice, but repeatable baselines save a lot of time.

[kernelrunner]

That matches what I was seeing. I ended up in about the same place using SGLang. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.